1. Field of the Invention
The present invention relates generally to a method and apparatus for multiple mode imaging, and more particularly to catadioptric optical systems used for dark field imaging applications.
2. Description of the Related Art
High precision optical instruments and imaging systems used in many different applications must operate effectively and efficiently. To accommodate optical functionality under varied conditions, precision lenses are often employed in different complex combinations.
Many different imaging modes exist for optical inspection. These imaging modes include bright field, confocal, and a variety of dark field imaging modes. Typically each different mode requires a different machine. Full inspection of an object, such as a semiconductor wafer, requires several separate very expensive machines. Combining many different imaging modes into one machine can dramatically reduce inspection costs as well as provide performance advantages.
The bright field imaging mode is commonly used in microscope systems. The advantage of bright field imaging is the image produced is readily distinguishable. The size of image features accurately represents the size of object features multiplied by the magnification of the optical system. This technique can be more easily used with image comparison and processing algorithms for computerized object detection and classification.
The confocal imaging mode has been successfully used for optical sectioning to resolve the height differences of object features. Most imaging modes have difficulty detecting changes in the height of features. The confocal mode forms separate images of object features at each height of interest. Comparison of the images then shows the relative heights of different features.
The dark field imaging mode has been successfully used to detect features on objects. The advantage of dark field imaging is that flat specular areas scatter very little light toward the detector, resulting in a dark image. Any surface features or objects protruding above the object scatter light toward the detector. Thus, in inspecting objects like semiconductor wafers, dark field imaging produces an image of features, particles, or other irregularities on a dark background.
Dark field illumination provides a large signal for small features that scatter light. This large signal allows larger pixels to be used for a given feature size, permitting faster object inspections. Fourier filtering can also be used to minimize the repeating pattern signal and enhance the defect signal to noise ratio.
Each dark field mode consists of a specific illumination scheme and collection scheme such that the scattered and diffracted light collected from the object provides the best signal. Several optical systems have been developed that use different dark field imaging modes including laser directional dark field, double dark field, and central dark ground.
One prior method for achieving laser directional dark field imaging is disclosed in U.S. Pat. No. 5,177,559, issued Jan. 5, 1993 to Batchelder and Taubenblatt and assigned to International Business Machines, which is hereby incorporated by reference. This method uses a collimated beam of monochromatic light to illuminate a semiconductor wafer from outside the objective between an angle of 8 degrees from the horizontal and the numerical aperture, or NA, defined by the imaging objective. Before forming a dark field image, the collected light passes through a Fourier filter to attenuate the spatial frequency components corresponding to repeating array patterns.
This laser directional dark field method illuminates the wafer outside the NA of the imaging objective. For this reason, the illumination angles are limited to between 8 degrees from the horizontal and the NA defined by the imaging objective. Collection angles are also limited to the range of angles within the NA of the objective. A long working distance objective is necessary to allow access by the laser to the area of interest on the semiconductor wafer. Objectives used in dark field applications of this type are generally limited to NAs less than 0.7, which corresponds to collection angles of only up to 44 degrees from normal.
Another prior method for achieving laser directional dark field imaging is disclosed in U.S. Pat. No. 5,428,442, issued Jun. 27, 1995 to Lin and Scheff and assigned Optical Specialties, which is hereby incorporated by reference. This method uses a collimated beam of monochromatic light illuminating the wafer from inside the optical system, within the NA defined by the objective. If the system will encounter a specific range of defect sizes, the illumination angle on the wafer is chosen so the optical system collects spatial frequencies of interest.
This is a laser directional dark field method wherein the laser illuminates the wafer from inside the NA as defined by the objective. The system uses the same objective pupil plane for injecting the illumination and processing the light collected by the objective. This objective pupil feature seriously limits the types of illumination and Fourier filtering that are possible. Systems using objectives of this type are generally limited to NAs of less than approximately 0.9. This means illumination angles are limited to less than approximately 64 degrees. Illumination at angles above 64 degrees is often necessary to obtain optimum defect sensitivity. This high angle illumination is not possible without a higher NA objective. The available objectives of this type with a high NA have very small fields, relative to that of a lower NA objective. This seriously limits the number of resolvable points in the image and the achievable inspection speed. Another problem with this technique is small amounts of scattered and reflected light from lens elements in this system have the ability to produce noise at levels that compromise the sensitivity. Introducing laser illumination from inside this type of objective can cause a significant amount of scattered and reflected light from the multiple lens surfaces. The system must deal with scattered light from the lenses, illumination beam and the specular reflection from the wafer, which is a tremendous potential problem.
A third known dark field imaging method designed to detect particles on a periodic patterned object is disclosed in U.S. Pat. No. 4,898,471, issued Feb. 6, 1990 to Stonestrom et al. and assigned to Tencor Instruments, which is hereby incorporated by reference. This method uses a single light beam scanned at a shallow angle over the object. The position of the collection system as well as the polarization of the light beam may be arranged to maximize the particle signal compared to the patterned signal.
This single beam/shallow angle system uses an off axis collector to image the area of interest onto a detector. The position of the collection system as well as the polarization of the illumination is arranged to maximize the signal scattered by particles. Only a single angular position is used for illumination and another angular position for collection. Such a dark field system uses a single spot to scan across the wafer in conjunction with a single detector. If the system uses a small spot for high sensitivity detection, the inspection speed tends to decrease dramatically. If a larger spot size is used to increase inspection speed, the overall system sensitivity degrades.
In the practical industrial application for object inspection the scattered and diffracted light is collected from either side of the plane of incidence. Since this dark field mode collects light outside of the plane of incidence, this mode is categorized as double dark field. The double dark field technique often obtains maximum sensitivity when the collection angle is greater than 70 degrees from normal, which is well outside the range of a 0.9 NA objective. This makes the combination of the double dark field mode and other imaging modes such as bright field and laser directional dark field difficult.
Three physical embodiments for performing dark field imaging are presented in FIGS. 1-3. FIG. 1 presents a directional dark field system 100 which illuminates the object 101 using a laser beam 102 directed at a high angle of incidence. Light contacting an anomaly is scattered or diffracted upward through collector 103, lens 104, and finally to the image plane 105.
FIG. 2a is side view of a double dark field design system 200 which illuminates the object 201 using laser beam 202 directed at a relatively low angle of incidence. Collectors 203 and 204 are mounted at different angles from the laser beam 202, typically 90 degrees. FIG. 2b illustrates a top view of the system with the collectors 203 and 204 mounted 180 degrees from one another and 90 degrees from the laser 202. Variations of these angles are possible. This provides enhanced collection capability and allows detection of particular object faults.
FIG. 3a illustrates a variation of a central dark ground imaging system 300, wherein the laser beam 301 passes through the collector 302 at an approximately perpendicular angle to the object 303. The light beam strikes the object and is diverted, depending upon the features encountered, toward various collectors mounted about the object. Four collectors 305-308 in FIG. 3b have been employed in the past, each at an angle 90 degrees from the nearest collectors, as shown in FIG. 3b. Different numbers of collectors may be used at various angles depending upon the type of object scanned and the defects anticipated.
The drawbacks of these prior dark field systems are twofold. First, physical limitations tend to restrict the angle at which light beams may be applied to the object, i.e. at either extremely deep or extremely shallow angles. This angular limitation tends to impede a full examination of the object. For example, if a type of defect may be detected best when light is applied at a 45 degree angle, such defects may not be apparent when the illumination source is limited to an angle less than 45 degrees. Secondly, these systems only employ one dark field mode each, which again tends to limit the types of defects which may be detected. Any attempt to combine different dark field imaging systems would be severely inhibited due to physical component placement restrictions and the inherent cost associated with multiple components.
With respect to the use of a single imaging mode, different schemes have varying performance advantages. For example, a double dark field arrangement is useful when small particles exist on the object, or rough films are expected to be encountered. Microscratches are best detected using normal incidence illumination and off normal imaging. Grazing incidence and back scattering may be used to detect missing contacts or vias on semiconductor wafers, while full sky mode best address metal grain variations. As the types of defects which may be encountered vary depending on many manufacturing factors, it would be best to inspect object for all of these defects. The problem, however, is that no single mode provides these capabilities, nor is a combination of these modes physically realizable nor economically feasible. Further, the use of multiple arrangements running over different machines negatively effects the scanning, processing, and evaluating time for each specimen.
It is therefore an object of the present invention to provide an apparatus that can incorporate as many imaging modes as possible and still maintain image quality and throughput requirements. Such an apparatus would be based on a high NA catadioptric objective with highly flexible illumination and imaging capabilities. Several prior catadioptric systems have been developed for microscopic inspection. While these systems may have a similar appearance to the present invention, they simply do not have the capability to integrate multiple inspection modes.
One apparatus previously used for microscopic inspection is the catadioptric bright field imaging system disclosed in U.S. Pat. No. 5,031,976, issued Jul. 16, 1991 to Shafer and assigned to KLA Instruments, which is hereby incorporated by reference. The system employs a focusing lens, a field lens, and a spherical mirror to focus and reflect light toward an object and back through an aperture located in the spherical mirror. The broadband aspects of the system disclosed do not directly translate to dark field imaging, but the patent does present an apparatus capable of 0.6 NA broad-band UV imaging for the deep ultraviolet spectral region, i.e. approximately 0.19 to 0.30 micron wavelengths.
The '976 patent is based on the Schupmann achromatic lens principle, which produces an achromatic virtual image, wherein an achromatic real image is produced from the virtual image using a reflective relay. The lens system is formed from a single glass type, fused silica. A series of lenses and mirrors is used to correct for aberrations and focus the light.
The system used in the '976 patent is shown in FIG. 4. The system includes an aberration corrector lens group 401 for correcting image aberrations and chromatic variation of the image aberrations, a focusing lens 403 for receiving light from the group 401 and producing an intermediate image plane at plane 405, a field lens 407 of the same material as the other lenses placed at the intermediate image plane 405, a thick lens 409 with a plane mirror back coating 411 whose power and position are selected to correct the primary longitudinal color of the system in conjunction with the focusing lens 403, and a spherical mirror 413 located between the intermediate image plane and the thick lens 409 for producing a final image 415. Most of the focusing power of the system is due to the spherical mirror 413 which has a small central hole 417 near the intermediate image plane 405 to allow light from the intermediate image plane 405 to pass through to the thick lens 409. The mirror coating 411 on the back of the thick lens 409 also has a small central hole 419 to allow light focused by the spherical mirror 413 to pass through to the final image 415.
The '976 system is a catadioptric objective capable of 0.6 NA, with broad-band UV imaging over a 0.5 mm field. This limited NA and field size does not support high speed inspection with more than one dark field mode. An NA of 0.6 collects only 20% of the available solid angle above the sample object. The maximum field size for this system is 0.5 mm. This limits its maximum pixel size and data collection speed. For example, a system with 0.5 mm field size would only have 25% of the maximum pixel size and data collection speed compared to a system with a 2 mm field size. The thickness of the lens mirror element in the '976 patent prevents the system from being scaled up for larger field sizes. The objective is formed from fused silica and uses a positive fused silica field lens near the intermediate image to correct for low order lateral color. The thick lens mirror element is also used in the catadioptric group to compensate the chromatic variation in aberrations of the external lens group. This broad band objective also has no provision for laser illumination through the catadioptric elements.
U.S. Pat. No. 5,488,229 to Elliott and Shafer and the associated, continuation in part application, both of which are hereby incorporated by reference, provide a modified version of the optical system of the '976 patent, one which has been optimized for use in 0.193 micron wavelength high power excimer laser applications such as ablation of a surface 421′ as seen in FIG. 5. This system has an aberration corrector group 401′, focusing lens 403′, intermediate focus 405′, field lens 407′ thick lens 409′, mirror surfaces 411′ and 413′ with small central opening 417′ and 419′ therein and a final focus 415′ as in the prior '976 patent, but repositions the field lens 407′ so that the intermediate image or focus 405′ lies outside of the field lens 407′ to avoid thermal damage from the high power densities produced by focusing the excimer laser light. Further, both mirror surfaces 411′ and 413′ are formed on lens elements 408′ and 409′. The combination of all light passing through both lens elements 408′ and 409′ provides primary longitudinal color correction of the single thick lens 409 in FIG. 4, but with a reduction in total glass thickness. Since even fused silica begins to have absorption problems at the very short 0.193 micron wavelength, the thickness reduction is advantageous at this wavelength for high power levels. Though the excimer laser source used for this optical system has a relatively narrow spectral line width, the dispersion of silica near the 0.193 micron wavelength is great enough that some color correction is still needed. Again, the system has a numerical aperture of about 0.6.
U.S. Pat. No. 5,717,518 to Shafer, Chuang, and Tsai and the associated continuation-in-part application describe a catadioptric objective capable of high NA, ultra broad-band UV bright field imaging. This is a 0.9 NA system with very tight manufacturing tolerances, primarily due to the broad-band requirements of the design. The broad-band UV nature of the design is due in part to the use of two glass types, calcium fluoride and fused silica. Calcium fluoride has excellent UV transmission properties but it is not necessarily a desirable glass because it is difficult to polish, not mechanically stable, and expensive. Fused silica is a more desirable material because in addition to its good UV transmission it has desirable manufacturing properties.
The system in the '518 patent is shown in FIG. 6. The focusing lens group 11 consists of seven elements 21-27 with two of the lenses 21 and 22 separated from the remaining lens elements. These two lenses form a nearly zero power doublet for the correction of chromatic variation of monochromatic imaging aberrations. The five lenses 23-27 form the main focusing subgroup. The subgroup focuses the light to an intermediate image 13. The curvature and positions of the lenses in this subgroup are selected to minimize the monochromatic aberrations as well as cooperate with the doublet 21-22 to minimize chromatic variations of those aberrations. The field lens group 15 typically composes an achromatic triplet. Both fused silica and calcium fluoride glass materials are used. The catadioptric group 17 includes a meniscus lens 39 with a concave reflective surface coating 41. It also includes a second optical element consisting of a lens element 43 with a reflective surface coating 45 on a back surface of the lens. The first lens 39 has a hole 37 centrally formed along the axis of the optical system. Light from the intermediate image 13 passes through the optical aperture 37 in the first lens 39 and then throughout the body of the second lens 43 where it is reflected back by mirror surface 45. The light then returns through the body of 43 and the body of 39 to reflect off surface 41. The light, now strongly convergent, passes through the body of 43 and an aperture 47 in the reflective surface coating to form the final image 19.
Common problems with using several of these prior catadioptric optical systems for multiple imaging modes includes: these are complicated systems with tight manufacturing tolerances, they provide no illumination and imaging above 0.9 NA and a field of view less than 1 mm, pupil planes that are not easily accessible, and limited imaging modes and associated illumination schemes.
It is therefore an object of the present invention to provide a method and an apparatus which can incorporate as many imaging modes as possible while still meeting image quality and throughput requirements.
It is another object of the current invention to provide an effective, relatively simple and low cost imaging system having high optical performance and minimal inspection times.
It is a further object of the current invention to provide a high numerical aperture imaging system having high optical performance.
It is yet another object of the current invention to provide a system having a pupil plane at a location separate and apart from the lenses of the optical system, and providing the ability to magnify the resultant image over a range of magnification values.